AI, Copyright & Creativity — Part 1 of 4
Artificial intelligence can generate an image in seconds. Sometimes the result resembles an artist’s distinctive work. But how does AI learn to do this, and what does it mean for the people who made the original art?
How does AI learn?
Generative AI systems learn patterns from vast collections of material, which can include photographs, paintings and illustrations. They use those patterns to produce images in response to a request.
The concern begins when an artist’s copyrighted work is included in training without their knowledge or individual permission. Whether that use is lawful depends on the circumstances and the applicable copyright rules.
Inspiration or copying?
Artists have always learned from one another. Techniques and ideas travel between generations, and entire art movements grow from shared influences.
Copyright, however, makes an important distinction: it protects original artistic expression, not an artistic style or an idea on its own. Someone cannot own the concept of abstract painting, for example. Reproducing protected elements of a particular artwork can be a different matter.
AI makes that boundary harder to discuss, especially when generated images closely resemble existing work.
The important question is not simply whether AI should learn from art, but under what conditions that learning should take place.
Next Friday: What does European copyright law say about AI training?
Further reading: Belgian FPS Economy — What is protected by copyright?